Shiplog Shiplog Orchestrator

Session orchestration, progress tracking, and background agent management

install
source · Clone the upstream repo
git clone https://github.com/aadivar/shiplog
Claude Code · Install into ~/.claude/skills/
T=$(mktemp -d) && git clone --depth=1 https://github.com/aadivar/shiplog "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/shiplog" ~/.claude/skills/aadivar-shiplog-shiplog-orchestrator && rm -rf "$T"
manifest: skills/shiplog/SKILL.md
source content

Shiplog — Session Orchestration Skill

You have the Shiplog plugin active. This skill handles automatic project tracking, session orchestration, and background agent management.

Session Start Behavior

At the start of every session, you MUST:

  1. Read progress: Read
    docs/PROGRESS.md
    to understand the current project state
  2. Display status: Show a one-line status summary:
    Shiplog: Sprint {N} | Feature {ID} ({status}) | {done}/{total} features done
    
  3. Load context: Use the information from PROGRESS.md to understand:
    • What sprint we're on
    • What feature was last worked on
    • What was done in the last session
    • What's planned next

CRITICAL: Post-Commit Agent Launching

After every

git commit
you perform, you MUST check for
.shiplog/config.json
. If it exists, launch background agents immediately. Do NOT wait for hook output. Do NOT skip this step.

How to launch agents after a commit

  1. Read config: Read

    .shiplog/config.json
    to get agent model and which agents are enabled.

    • Handle both config formats:
      • Nested:
        { "agents": { "model": "sonnet", "specs": { "enabled": true } } }
      • Flat:
        { "agentModel": "sonnet", "agents": { "specs": true } }
  2. Read agent prompts: For each enabled agent, read its prompt file from the plugin's

    agents/
    directory:

    • agents/specs-agent.md
    • agents/prd-agent.md
    • agents/security-agent.md
    • agents/memory-agent.md
  3. Launch agents using the

    Agent
    tool with
    run_in_background: true
    and
    subagent_type: "shiplog:{agent}-agent"
    :

    • Specs Agent → updates
      docs/SPECS.md
    • PRD Agent → updates
      docs/PRD.md
      and
      docs/PROGRESS.md
    • Security Agent → updates
      docs/SECURITY.md
    • Memory Agent → updates memory files
  4. Launch in parallel: Launch specs, PRD, and security agents together. Memory agent can also run in parallel.

  5. Don't block the user: Launch all agents in background and continue immediately. Never wait for agents to complete before responding.

Fallback: Hook trigger

If you see

SHIPLOG_AGENTS_TRIGGER
output from a PostToolUse hook, treat it the same way — launch agents. But do NOT rely on this; always proactively launch after commits.

Custom Agents

If

.shiplog/agents/
directory exists in the project, also launch each custom agent found there.

Agent failure handling

  • If any agent fails, it fails silently — do not interrupt the user's work
  • Do not retry failed agents
  • Do not report agent results unless the user asks

Available Commands

  • /shiplog init
    — Initialize Shiplog on current project
  • /shiplog status
    — Show project status dashboard
  • /shiplog feature <name>
    — Create a new feature from template
  • /shiplog sprint <n>
    — Start a new sprint
  • /shiplog review
    — Manually trigger all background agents
  • /shiplog config
    — Show/edit configuration

File Locations

  • Config:
    .shiplog/config.json
  • PRD:
    docs/PRD.md
  • Specs:
    docs/SPECS.md
  • Security:
    docs/SECURITY.md
  • Progress:
    docs/PROGRESS.md
  • Features:
    docs/features/
  • Agent prompts:
    agents/
    (in plugin root)
  • Templates:
    templates/
    (in plugin root)